Data Scientist • ML Engineer • PhD Researcher
Hello! I'm a Data Science PhD candidate at the University of Oklahoma, passionate about building intelligent systems. My research spans Scientific Machine Learning, Open-World Classification, and time-series analysis. I specialize in creating interpretable AI models with real-world applications, from robotics to healthcare.
A framework that automates the discovery of nonlinear dynamics.
A hybrid framework that automates the discovery of nonlinear dynamics by combining symbolic regression and sparse identification. It generates basis functions automatically, recovering ground-truth equations with 92.8% accuracy even in noisy conditions, and outperforming traditional methods in prediction and simplicity.
An adaptive framework for identifying and categorizing previously unseen classes.
Built an adaptive framework for identifying and categorizing previously unseen classes in dynamic environments. Achieved 75-93% accuracy improvement across text, image, and sensor datasets, advancing beyond traditional closed-world assumptions.
An algorithm that incorporates graph-informed regularization for the discovery of network dynamics.
Developed a novel algorithm incorporating graph-informed regularization into sparse regression for accurate discovery of network dynamics. Validated on Stuart-Landau oscillator networks with superior performance in identifying complex system behaviors.
A framework for extracting EEG features to improve Alzheimer's detection.
Developed multilayer cross-frequency functional connectivity framework for extracting spatiotemporal EEG features. Achieved 15-20% accuracy improvement over baseline methods for early Alzheimer's disease detection.
A project that involved engineering a relational database structure and optimizing it.
Engineered a relational database structure, optimizing data storage and indexing strategies (Azure SQL). Constructed a database application to execute complex SQL queries for efficient transaction management (Java, SQL).
A deep learning face recognition system for time-clock systems and service robots.
Architected and deployed a deep learning face recognition system achieving 95-97% accuracy under challenging real-world conditions. Implemented for time-clock systems and service robots with robust data preprocessing pipelines.
A project that involved creating path-planning algorithms for obstacle avoidance.
Created physics-informed and ML-based path-planning algorithms with novel loss functions for superior obstacle avoidance. Modeled aerial robot formation control with security measures against cyber-physical attacks.
An interactive web application that calculates the optimal leveling strategy.
Interactive web application implementing greedy algorithm optimization for resource management. Calculates optimal leveling strategy to minimize coin expenditure while maximizing point gains using efficiency-based decision making.
A collection of various machine learning and data analysis projects.
A collection of projects including a movie database console app, predictive analysis of cancer gene expression, credit card fraud detection, HR analytics on employee attrition, and student performance analysis.
University of Oklahoma
University of Oklahoma | Data Science & Analytics Institute
Tosan Techno Company & TaarLab
University of Tehran
TaarLab Human and Robot Interaction Laboratory
Intelligent Control Systems Laboratory
University of Oklahoma
University of Tehran
Amirkabir University of Technology
I'm currently pursuing my PhD at the University of Oklahoma and always interested in research collaborations, innovative projects, and opportunities in AI/ML. Let's connect!